Portrait profile for P30+/P45+

[torger]

Hi Edmund,

Would it be used with 'film curve' or 'linear response'?

A quick comment on this since I've recently worked with this -- you should not use a capture one ICC profile with a radically different curve than it was designed for. The various contrasty curves are close enough to the default film curve, but the linear curve is quite far off which can lead to distortions of the color, in the tests I've made you often get a bit greenish cast of darker skin tones.

The reason for this is that LUT-based profiles have different color corrections depending on brightness. Eg a bright blue may get a different hue/saturation adjustment than a dark blue. The contrasty curves brightens the image compared to the linear curve. As the ICC profile is applied *after* the curve is applied this means that you'll get different color adjustments depending on the curve.

If we look at DNG and DCP many profiles there are "2.5D" ie they have only hue and saturation correction which is the same regardless of brightness. This is nice because they get brightness-independent. However to get the best corrections one need to be able to have different corrections depending on color brightness and DCP supports this to with full 3D profiles. DCP profiles is however applied before the tone curve which may seem good (as it would be "curve-independent"), but it should not not make any real difference in practice as you always need to have some sort of curve target you calibrate with, and I assume Adobe's profiles are designed for their default curves. Adobe also applies an S-type of contrast curve which causes more color shifts than the gamma-like curve Capture One generally uses. I should say though that I haven't investigated color shifts in Adobe's DCPs in relation to the curve as much as I've done with Capture One's ICC profiles, but I would expect that you'd have same type of performance issues with Adobe's profiles if you change curve to a linear curve (I don't remember now if it's even possible to assign a linear curve in Lightroom though).

[yaya]

The linear curve in Capture One was designed to be used for reproduction, where the end user will often create their own input profiles. Note that it is intended for use during the capture stage (typically tethered) and the lights should be adjusted accordingly. If one tries to use in in post the result will be under exposed and the tones will be "squeezed"

Yair

[eronald]

I'd like to remind people that they are free to download my old profile and make up their own minds.
Think of it like trying out a new film, except in this case there is no need to go out and shoot.
I recommend the standard film-like C1 curve, but of course anyone is free to experiment.

Edmund

[TMARK]

I used to, on occasion, shoot heavily lit fashion tethered to C1, using the linear curve.  I found it easier to create contrast with light.  This was mainly shooting tests for models and testing lighting set ups.  The C1 Film standard could often result is crushed blacks and blown highlights if you light a set using a meter only.

[Iliah]

Dear Erik,

> Please recommend a calibration software, that generates ICC profiles for C1 from a calibration card

Can you upload a couple of your shots of calibration targets, raw?

[ErikKaffehr]

Hi,

But I am on travel, back on thursday.

Best regards
Erik

Dear Erik,

> Please recommend a calibration software, that generates ICC profiles for C1 from a calibration card

Can you upload a couple of your shots of calibration targets, raw?

[ErikKaffehr]

Hi,

Here they are: http://echophoto.dnsalias.net/ekr//Articles/FakeProfiles/CalibShots/

The ColorChecker shot is what I use for all my profiles. On the Sony I am quite happy with the built in profiles, so I did not mess around with making my own.

Best regards
Erik

Dear Erik,

> Please recommend a calibration software, that generates ICC profiles for C1 from a calibration card

Can you upload a couple of your shots of calibration targets, raw?

[Iliah]

Dear Erik,

I'm afraid the shot is not quite good enough for profiling, the values on the black frame around ColorChecker Passport vary up to 12%, and the white balance is not uniform enough too. I placed 66x66 pixels samples on the frame like this, numbered clockwise, first one is upper left corner.

Exported values to csv, original csv attached. Opened the file in Excel and for each colour channel calculated sample average divided by minimum average, Excel spreadsheet attached.
On a side note, I would try to clean the target or, even better, use a ColorChecker SG as it in my experience always result in more realistic and usable profiles compared to using just ColorChecker.

[eronald]

I do of course regret that when asked to contribute to the specs for this target I didn't more vocally insist on a uniformity check. Mea culpa. Maybe we could debate on features we could request for a future version? My feeling is that the smaller the total sample area the better, as this reduces light falloff effects  across the active area, but some tricks might improve software flare reduction too.

Edmund

Dear Erik,

I'm afraid the shot is not quite good enough for profiling, the values on the black frame around ColorChecker Passport vary up to 12%, and the white balance is not uniform enough too. I placed 66x66 pixels samples on the frame like this, numbered clockwise, first one is upper left corner.

Exported values to csv, original csv attached. Opened the file in Excel and for each colour channel calculated sample average divided by minimum average, Excel spreadsheet attached.
On a side note, I would try to clean the target or, even better, use a ColorChecker SG as it in my experience always result in more realistic and usable profiles compared to using just ColorChecker.

[Iliah]

Yes, flare is a problem. For uniformity, maybe an N8 grey border? The cross in the middle is unnecessary IMHO, as in my experience it is better to shoot targets out of focus.

[eronald]

Yes, flare is a problem. For uniformity, maybe an N8 grey border? The cross in the middle is unnecessary IMHO, as in my experience it is better to shoot targets out of focus.

The idea was to provide an aim point, and an element of repeatability. Otherwise people tend to choose one of the fields but not always the same one as an aim point. What was the name of that book? Camera Profiling for Dummies?

Edmund

[Iliah]

I see. What I would suggest is a neutral grey cover sheet for flat field and an overlay for any focusing needs.

[eronald]

I see. What I would suggest is a neutral grey cover sheet for flat field and an overlay for any focusing needs.

What features would you like to see adjacent to the color fields of a chart?

Edmund

[Iliah]

If there is a hard N8 sheet to cover the target, all I want that the plastic frame has no specular reflections.

[eronald]

I'm doing a different design

What sort of fields would you like on the target?

Please remember that when doing "Profiling for Dummies" you basically get one test shot. Which should be sharp because in a repro cycle it is a production exposure.

Edmund

[ErikKaffehr]

Dear Iliah.

Learning all the time ;-)

I would expect the target to be evenly illuminated, as all corners are approximately equal distance from the sun, but I was concerned about contamination from surrounding light. This time I had my target on metal table.

Can we use studio flash for D50 or D60 reference or we do need real sunlight? Studio flash is simpler, but I have not really found a description of spectral distribution of flash light, so I don't know how well it approximates an D-series illuminant.

Anyway, one nice thing with raw is you can always change a profile.

Best regards
Erik

Dear Erik,

I'm afraid the shot is not quite good enough for profiling, the values on the black frame around ColorChecker Passport vary up to 12%, and the white balance is not uniform enough too. I placed 66x66 pixels samples on the frame like this, numbered clockwise, first one is upper left corner.

Exported values to csv, original csv attached. Opened the file in Excel and for each colour channel calculated sample average divided by minimum average, Excel spreadsheet attached.
On a side note, I would try to clean the target or, even better, use a ColorChecker SG as it in my experience always result in more realistic and usable profiles compared to using just ColorChecker.

[Vladimirovich]

This time I had my target on metal table.

what about a lot of light being reflected from metal surface back into your lens ?

[Vladimirovich]

Can we use studio flash for D50 or D60 reference or we do need real sunlight?

as far as I remember what he was writing about a simple setup - it was a stabilzed (voltage) halogen floodlight w/ gel(s), reflected off brushed aluminium metal surface on the target (not a direct illumination)

[ErikKaffehr]

Iliah is right.

Best regards
Erik

as far as I remember what he was writing about a simple setup - it was a stabilzed (voltage) halogen floodlight w/ gel(s), reflected off brushed aluminium metal surface on the target (not a direct illumination)

[ErikKaffehr]

Hi,

I don't think so. The metal area is dark, so it is reflecting away from the lens. But I think it may reflect skylight on the target. On the other hand, the grey patches had quite normal colour temperature, so I assumed that would not be a problem.

Lot of food for thought…

Best regards
Erik

what about a lot of light being reflected from metal surface back into your lens ?